1Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Screening and assessment tools for early detection of malnutrition in hospitalised children: a systematic review of validation studies

Petra Klanjsek,1 Majda Pajnkihar,1 Natasa Marcun Varda,2,3 Petra Povalej Brzan1,3,4

To cite: Klanjsek P, Pajnkihar M, Marcun Varda N, et al. Screening and assessment tools for early detection of malnutrition in hospitalised children: a systematic review of validation studies. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

► Prepublication history and additional material for this paper are available online. To view these files, please visit the journal online (http:// dx. doi. org/ 10. 1136/ bmjopen- 2018- 025444).

MP and NMV contributed equally.PK and PPB contributed equally.

Received 15 July 2018Revised 21 February 2019Accepted 27 February 2019

1Faculty of Health Sciences, University of Maribor, Maribor, Slovenia2Department of Paediatrics, University Medical Centre Maribor, Maribor, Slovenia3Faculty of Medicine, University of Maribor, Maribor, Slovenia4Faculty of Electrical Engineering and Computer Science, University of Maribor, Maribor, Slovenia

Correspondence toAssistant Professor Petra Povalej Brzan; petra. povalej@ um. si

Research

© Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

AbstrACt Objective The aim of the present study was to identify all currently available screening and assessment tools for detection of malnutrition in hospitalised children, and to identify the most useful tools on the basis of published validation studies.Design Systematic review.Data sources PubMed, CINAHL and MEDLINE were searched up to October 2017.Eligibility criteria for selecting studies Studies in English that reported sensitivity, specificity and positive/negative predictive values (PPVs/NPVs) in the paediatric population were eligible for inclusion.Data extraction and synthesis Two authors independently screened all of the studies identified, and extracted the data. The methodological qualities of the studies included were assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 tool.results The 26 validation studies that met the inclusion criteria for this systematic review used eight screening and three assessment tools. The number of participants varied from 32 to 14 477. There was considerable variability in the chosen reference standards, which prevented direct comparisons of the predictive performances of the tools. Anthropometric measurements were used as reference standards in 16 of the identified studies, and full nutritional assessment in 5. The Pediatric Yorkhill Malnutrition Score (PYMS) screening tool performed better than Screening Tool for the Assessment of Malnutrition and Screening Tool for Risk On Nutritional status and Growth when compared in terms of anthropometric measurements, especially for body mass index (Se=90.9, Sp=81.9) and triceps skinfold thickness (Se=80.0, Sp=75.0). However, low PPVs indicated the problem of overprediction of positive cases, which was typical for all of the studies that used anthropometric measurements as the reference standard.Conclusions This systematic review identifies the need for definition of the gold standard for validation of screening tools. Anthropometry measurements using WHO or Centers for Disease Control and Prevention growth charts should be considered as the possible reference standard in future validation studies. We would recommend the use of PYMS for hospitalised paediatric patients without chronic conditions, in combination with full nutritional assessment.PrOsPErO registration number CRD42017077477.

IntrODuCtIOnOver the last decade, several studies have shown that the prevalence of malnutrition in hospitalised children varies from 6.1% to 55.6% worldwide.1–6 The importance of early detection of malnutrition in hospitalised paediatric patients has led to the development of several nutritional screening and assess-ment tools. Screening tools are designed to provide early identification of children at risk of nutritional impairment, and they have the potential to improve health outcomes and to reduce healthcare costs. All patients consid-ered at risk during such screening should be referred for nutritional assessment and possible intervention. However, currently, there is no consensus on the appropriate screening tool to identify these children who are at risk of developing malnutrition during hospitalisation.7–9

Due to the absence of a gold stan-dard,8 10–14 screening/assessment tools are

strengths and limitations of this study

► This systematic review was based on a compre-hensive search and includes a large number of screening/assessment tools for evaluation of the malnutrition risk in hospitalised children, along with their validation studies.

► Only the studies in English that reported sensitivity, specificity and positive/negative predictive values or data enabling manual calculation of them were included.

► The methodological quality of the validation studies included was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 tool.

► This systematic review highlights the heterogeneity of both the tools available and their validation stud-ies, along with the challenges that result from this heterogeneity.

► Although our search included multiple electronic da-tabases and grey literature, relevant data that have not been reported may be missed.

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

2 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

usually validated using the following reference stan-dards: dietetic/nutritional assessment; anthropometric measures, as defined by the WHO15 16 ; growth charts of the Centers for Disease Control and Prevention (CDC)17 ; national growth charts or Roher’s Ponderal Index for newborns.

The CDC and National Center for Health Statistics growth charts for the USA were released in the year 2000.18 The WHO released international growth charts for chil-dren up to 5 years of age in 2006. In addition, in 2007, the WHO developed a growth reference for school-age children and adolescents (≤19 years).19 According to the WHO and CDC recommendations, the WHO growth charts are more appropriate for children aged from 0 to 2 years. The methods used to create the CDC and WHO growth charts were similar for children from 2 to 5 years of age. The CDC growth charts can be used for children <5 years.20 The WHO growth charts had already been adopted in 125 out of 219 countries by the end of April 2011.21 National growth charts have also been used in some countries, or parts of countries, including China22 and the United Arab Emirates.23

Dietetic/nutritional assessment is the systematic process of collecting and interpreting information to make deci-sions on the nature and causes of nutrition-related health issues that affect an individual. Full nutritional assess-ments also include biochemical parameters.24 However, dietetic/nutritional assessments vary across different countries due to differences in educational standards.3

Anthropometric measures such as weight for age (WFA), height for age (HFA), weight for height (WFH) and body mass index (BMI) for age, as SD scores (ie, Z-scores) are usually used for identification of malnutri-tion.25 Malnutrition can be acute (ie, wasting) or chronic (ie, stunting). Moderate acute malnutrition is usually defined using WFH, as Z-scores between −3 and −2. A WFH Z-score less than −3 indicates severe acute malnutrition. Chronic malnutrition is defined using HFA, as Z-scores between −3 and −2 for moderate chronic malnutrition, and Z-scores less than −3 for severe chronic malnutrition. Furthermore, mid-upper arm circumference (MUAC) is used for the identification of malnutrition in infants and children aged 3 months to 5 years, with the cut-off Z-score of less than −2.25 26

The aim of the present study was to systematically review the available publications on the screening and/or assessment tools for hospitalised children, with a focus on the ability of these tools to predict the risk or presence of malnutrition, in order to identify the most useful tool for use in the clinical environment.

MEthODsDesignThis systematic review of published validation studies was registered with PROSPERO (online supplementary file 1). The findings are reported according to the guidelines of the Preferred Reporting Items for Systematic Reviews

and Meta-Analyses (PRISMA) statement27 (see PRISMA checklist; online supplementary file 2).

We focused on answering the following research ques-tions (RQs) (see online supplementary file 3): RQ1: What are the currently available screening and assessment tools for detecting malnutrition in hospitalised paediatric patients? RQ2: What is the validity of the screening and assessment tools versus the reference standard?

search strategyTo identify all relevant publications, we performed system-atic searches in the following bibliographic databases: PubMed, CINAHL and MEDLINE. The searches were carried out on October 20, 2017. The keyword combi-nations used included the following: premature*, imma-ture*, child*, baby, infant*, newborn*, neonate*, kid*, babies, adolescent*, pediatric*, paediatric*, screen*, assess*, tool*, undernutrition*, undernourish*, malnu-trition, malnourish*. The search strategies are outlined in online supplementary file 4, and they were adapted to each database and kept consistent across all searchers. The reference lists of the identified studies were manually searched for potentially relevant studies.

study selectionAll potentially relevant titles and abstracts were blinded for author, journal and year of publication, and then screened for eligibility by two reviewers (PK and PPB), independently. Differences in judgement were resolved through a consensus procedure. All of the studies obtained from the bibliographic databases were entered into EndNote X8, and duplicates were excluded. The inter-rater agreement between the reviewers based on Cohen’s kappa statistic was 0.79 for 576 studies. The full texts of the selected studies—no longer blinded to authors and journals—were obtained for further review by two reviewers, independently (PK and PPB), to judge for eligibility. The Cohen’s kappa coefficient here was 0.93 for 64 studies, which indicated a high level of agree-ment. In cases of doubt, a decision was made by a third reviewer (NMV).

The studies were reviewed to ensure their focus was aligned with the purpose of the literature review. Those that were clearly inappropriate to answer the RQs and/or did not fit the predefined inclusion criteria were excluded. The flowchart of the complete search and selection process is shown in figure 1.

Inclusion/exclusion criteriaThe studies eligible for inclusion were validation studies in English that reported sensitivity (Se), specificity (Sp) and positive/negative predictive values (PPVs, NPVs, respec-tively) in paediatric populations. All of the predefined inclusion and exclusion criteria were used as outlined in online supplementary file 5. The list of studies (n=38) not meeting the selection criteria after reading the full text is presented in online supplementary file 6.

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

3Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Quality appraisalThe methodological quality assessment of the studies was performed using Review Manager V.5.3,28 with a revised tool for the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2.29 The QUADAS-2 tool uses four key domains to rate the risk of bias and the applicability of primary diagnostic accuracy studies. The key domains are as follows: patient selection (sampling, inclusion/exclu-sion criteria, sampling bias, adequacy), index tests (the validated tool, correct use and interpretation, possible bias), reference standards (the reference tool, correct use and interpretation, possible bias) and the flow and timing (the sequence, time interval, correct performance of reference standard and index test, possible bias). The results from QUADAS-2 can be expressed as high/unclear/ low risk of bias and as high/unclear/ low appli-cability concerns.

Data extraction and synthesisThe data were extracted by two reviewers (PK and PPB) and checked by a third reviewer (MP) using the predefined data-extraction criteria, which included the following: authors and country, nutritional screening/assessment tool used, study type, sample size, age of partic-ipants and reported clinical performance. To evaluate

the clinical performance and diagnostic accuracy of the screening tools, the following criteria were considered: Se, Sp, PPV and NPV. Studies that did not report Se, Sp, PPV and/or NPV, but that provided the data that enabled calculation of these values, were also included in the study. These metrics were subsequently calculated manu-ally by the authors and are indicated as such in the results tables.

The validation of the reproducibility and reliability of the screening/assessment tools was also considered, using data from the agreement analysis between the assessed tools and the chosen reference, as well as the inter-rater agreement shown in the studies.

For reasons of clarity, we have rated the results of each study as good, moderate/fair or poor validity. The kappa values were rated by the classification system proposed by Landis and Koch.30 Although the literature does not provide general cutoffs for Se and Sp, as they greatly depend on the clinical consequences, we have rated the values to maintain transparency and clarity, as in van Bokhorst-de van der Schueren et al.31 All of the cutoff points are described in table 1.

Patient and public involvementNo patients or public were involved in the present study.

Figure 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of the search and study selection process.

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

4 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

rEsultssearch resultsFigure 1 shows the flow diagram of the search and study selection process. We identified 724 studies initially, of which 26 met all of the pre-established eligibility criteria and were included in the critical appraisal.

Paediatric nutritional screening tools identifiedDuring this systematic review, we identified eight vali-dated pediatric nutritional screening tools and three nutritional assessment tools for hospitalised paediatric patients. These can be classified based on their special-ties, as follows:

► Medical and Surgical Department: Screening Tool for the Assessment of Malnutrition in Paediatrics (STAMP),32 33 Paediatric Yorkhill Malnutrition Score (PYMS),7 34 Screening Tool for Risk On Nutritional status and Growth (STRONGkids)

2 and Paediatric Nutrition Screening Tool (PNST).9

► Surgical Department: Subjective Global Assessment (SGA)35–38 and Subjective Global Nutritional Assess-ment (SGNA).13

► Oncology Department: Nutrition Screening tool for childhood Cancer (SCAN),39 which was developed specifically for children with cancer.

► Pulmonology Department: Nutrition screening tool for pediatric patients with cystic fibrosis (CF),40 and Nutritional risk screening tool in CF41 for pediatric patients with CF.

► Neonatal Intensive Care: Neonatal Nutrition Screening Tool (NNST)42 for infants in the Neonatal Intensive Care Unit.

► Clinical Assessment of Nutritional Status (CANS) score,43 to differentiate malnourished from appropri-ately nourished babies.

However, six additional screening tools were identified in the eligibility step of PRISMA, although the associated studies were excluded in the final step as the inclusion criteria were not met:

► Medical Department: Nutrition Risk Score (NRS).44

► Medical and Surgical Department: Simple Paediatric Nutrition Risk Score (SPNRS)45 and Pediatric Digital Scaled MAlnutrition Risk screening Tool (PeDiS-MART).46 47

► Psychiatric Department: St Andrew's Nutrition Screening Instrument (SANSI).48

► Screening tool to predict malnutrition among chil-dren under 2 years old in Zambia.49

► Nutrition screening for infants and young children with special health care needs: A Look at Your Child’s Nutrition.50

Characteristics of the studies included in the systematic reviewThe characteristics of the 26 studies included in this systematic review are outlined in table 2. Sample sizes varied from 32 to 14 477 participants. Eleven studies (42.3%) excluded patients with length of hospital stay (LOS) of <24 hours.2 9 14 32 39 51–56 The studies often excluded children <1 year (57.7%),2 6 7 32 38–41 52 53 55–59 intensive care unit (ICU) patients (34.6%)2 14 38 51 52 58 60–62 and patients with unstable/specific conditions, such as oncology patients,53–55 61 conditions that affected hydra-tion,9 39 cardiology, renal and orthopaedic specialties,7 fever, diarrhoea,6 obesity53 and others. Some of the studies included only patients with particular conditions, where a specially designed screening/assessment tool was usually used. However, STAMP was designed for clinical and surgical patients and was validated also on patients with spinal cord injury (SCI)6 14 and with inflammatory bowel disease (IBD).59 Additionally, STAMP was also used in outpatients in two studies.57 59 The SGA assessment tool was originally designed for adults; however, in one of the selected studies, it was tested on children.38

There was relatively high heterogeneity in the choice of the reference standards. Anthropometric measurements were used in 18 of the selected studies (69.2%)2 6 9 38 42 51–55 58–65 and full dietetic/nutritional assessments were used in 5 studies (19.2%).7 14 32 41 57 Three studies (12.5%) used SGNA as the reference standard.9 39 56 The Cystic Fibrosis Foundation (CFF) Consensus Report criteria were used as the reference standard in both of the studies on patients with CF.40 41 Validation with other screening tools along with the reference standards were reported in four studies (16.7%).14 40 59 63 Nine studies (37.5%) reported validation of two or more screening tools with the same reference standard and on the same patients.9 53 54 56 57 59 61 63 65

risk of bias and applicability concernsThe results of the quality appraisal analysis using the QUADAS-2 tool are presented in online supplementary file 7.

The patient selection was considered as high risk or unclear because of the non-specific description of the patient selection process in three studies.7 41 58 Possible bias from conducting non-blinded index tests with respect to the results of the reference standard

Table 1 Cutoffs applied to assess the validity of the nutritional screening and monitoring tools

Assessment Code Rating Cutoff

Sensitivity (Se)/ g Good Se and Sp≥80%

Specificity (Sp) f Fair Se or Sp<80%, but both >50%

p Poor Se or Sp≤50%

Kappa30 pe Almost perfect 0.81–1.00

su Substantial 0.61–0.80

m Moderate 0.41–0.60

f Fair 0.21–0.40

s Slight 0–0.20

n No agreement <0

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

5Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Tab

le 2

C

hara

cter

istic

s of

the

stu

die

s in

clud

ed in

the

sys

tem

atic

rev

iew

Stu

dy

Too

lC

oun

try

Tim

e fr

ame

Po

pul

atio

nS

amp

le

size

Ag

e ra

nge

Exc

lusi

on

crit

eria

Ref

eren

ce s

tand

ard

Cho

urd

akis

et

al63

PY

MS

; STA

MP

; S

TRO

NG

kid

s

12 c

ount

ries:

C

roat

ia,

Isra

el,

Den

mar

k,

Italy

, Fra

nce,

P

olan

d,

Ger

man

y,

Gre

ece,

The

N

ethe

rland

s,

UK

, Rom

ania

Feb

201

0 to

Ju

ly 2

011

Clin

ical

and

sur

gica

l p

atie

nts

2567

1 m

onth

to

18 y

ears

Acc

iden

t an

d E

mer

genc

y D

epar

tmen

t of

Day

Car

e U

nit

WH

O (B

MI,

MU

AC

, TS

FT)

Dur

akb

aşa

et a

l60S

TRO

NG

kid

sTu

rkey

Ap

ril t

o Ju

ly

2012

Sur

gica

l pat

ient

s49

4>

1 m

onth

Age

 <30

day

s, a

dm

itted

to

clin

ics

othe

r th

an p

aed

iatr

ic

surg

ery

war

d o

r ha

d a

noth

er

oper

atio

n in

the

pre

ced

ing

30 d

ays

WH

O (W

FH, B

MI)

for

acut

e m

alnu

triti

on, W

HO

(HFA

) for

ch

roni

c m

alnu

triti

on

Gal

era-

Mar

tinez

et

al61

STR

ON

Gki

ds;

STA

MP

Sp

ain

May

201

3C

linic

al a

nd s

urgi

cal

pat

ient

s22

3>

1 m

onth

Pae

dia

tric

ICU

, onc

olog

y, d

ay

surg

ery

war

d, p

atie

nts

who

co

uld

not

be

wei

ghed

and

m

easu

red

at

adm

issi

on

WH

O (B

MI S

DS

)

Ger

asim

idis

et

al7

PY

MS

UK

23 J

une

to 2

8 O

ct 2

008

Clin

ical

and

sur

gica

l p

atie

nts

247

1–16

yea

rsC

ard

iolo

gy, r

enal

, ort

hop

aed

ic

spec

ialti

es, c

ritic

al c

are

Full

die

tetic

ass

essm

ent:

d

ieta

ry h

isto

ry, a

nthr

opom

etric

m

easu

rem

ents

. Bod

y co

mp

ositi

on

char

acte

ristic

s

Hul

st e

t al

2S

TRO

NG

kid

sTh

e N

ethe

rland

s26

–28

Nov

20

07C

linic

al a

nd s

urgi

cal

pat

ient

s41

7>

1 m

onth

to

18 y

ears

ICU

, LO

S <

1 d

ayW

HO

(HFA

, WFH

) on

The

Net

herla

nds

Nat

iona

l Sta

ndar

ds16

Huy

sent

ruyt

et

al51

STR

ON

Gki

ds

Bel

gium

Dec

201

0 to

A

pril

201

1C

linic

al a

nd s

urgi

cal

pat

ient

s36

81

mon

th t

o 16

yea

rs<

24 h

ours

LO

S, I

CU

, rea

dm

itted

fo

r sa

me

cond

ition

with

in 7

day

sW

HO

(WFH

) for

acu

te m

alnu

triti

on,

WH

O (H

FA) f

or c

hron

ic

mal

nutr

ition

John

son

et a

l42N

NS

TU

K20

10N

eona

tal p

atie

nts

in

ICU

909

Infa

nts 

<2

wee

ks a

t d

isch

arge

WH

O (W

FA) w

ith U

K-W

HO

gro

wth

ch

art

Ling

et

al52

STA

MP

; S

TRO

NG

kid

s

UK

Two

cons

ecut

ive

day

s in

A

ugus

t 20

09

Inp

atie

nts

566

wee

ks t

o 16

yea

rsA

ge <

1 m

onth

, LO

S <

1 d

ay,

heig

ht m

easu

rem

ent

not

accu

rate

ly o

bta

ined

, IC

U, d

ay

surg

ery

war

d

WH

O (W

FH, B

MI i

f HFA

); W

FH: h

eigh

t<12

0 cm

; BM

I: he

ight

>12

0 cm

; WH

O (H

FA) f

or

chro

nic

mal

nutr

ition

Mah

dav

i et

al38

SG

AIr

anJu

ne 2

008

to

Aug

200

8C

linic

al a

nd s

urgi

cal

pat

ient

s14

02–

12 y

ears

Em

erge

ncy

Dep

artm

ent,

ne

wb

orn

ICU

, new

bor

n S

pec

ial

Car

e U

nits

Ant

hrop

omet

ric, b

ioch

emic

al

mea

sure

men

ts (W

FA, H

FA,

WFH

, TS

F, M

AC

, ser

um a

lbum

in,

tran

sfer

rin).

Pat

ient

s cl

assi

fied

as

und

erno

uris

hed

whe

n at

leas

t tw

o p

aram

eter

s ar

e su

bno

rmal

Măr

gine

an e

t al

53S

TRO

NG

kid

s; m

odifi

ed

STR

ON

Gki

ds

Rom

ania

1 M

ay 2

011

to

30 J

an 2

012

Tert

iary

pae

dia

tric

te

achi

ng h

osp

ital

326

1–17

yea

rsO

bes

e, a

ctiv

e m

alig

nanc

y,

LOS

 <24

hou

rsW

HO

(WFH

, HFA

); to

tal s

erum

p

rote

in (n

orm

al: 6

.6–8

.7 m

g/d

L)

Con

tinue

d

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

6 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Stu

dy

Too

lC

oun

try

Tim

e fr

ame

Po

pul

atio

nS

amp

le

size

Ag

e ra

nge

Exc

lusi

on

crit

eria

Ref

eren

ce s

tand

ard

Mar

tinez

-Nad

al e

t al

64C

AN

S s

core

Sp

ain

2003

–201

4N

eona

tal p

atie

nts

14 4

77N

ewb

orns

Not

rep

orte

dR

oher

’s P

ond

eral

Ind

ex (P

I<2.

2 g/

cm3 )

McC

arth

y et

al32

STA

MP

UK

Ap

ril t

o Ju

ne

2004

Clin

ical

and

sur

gica

l p

atie

nts

238

2–17

yea

rsLO

S <

24 h

ours

, uno

bta

inab

le

wei

ght

or h

eigh

t m

easu

rem

ents

Full

nutr

ition

al a

sses

smen

t:

hist

oric

al r

ecor

ds,

nut

ritio

nal

mea

sure

men

ts, a

nthr

opom

etry

McD

onal

d41

Nut

ritio

n R

isk

Scr

eeni

ng T

ool

in C

F

US

AN

ot r

epor

ted

Pae

dia

tric

pat

ient

s w

ith 

CFd

iagn

osis

852–

20 y

ears

Not

rep

orte

dC

FF C

onse

nsus

Rep

ort

crite

ria;

full

die

tetic

ass

essm

ent:

MU

AC

an

d/o

r TS

F le

ss t

han

25th

p

erce

ntile

for

sex

and

age

, su

bop

timal

ser

um v

itam

in a

nd/

or t

race

min

eral

leve

ls, p

ulm

onar

y ex

acer

bat

ion,

sub

optim

al d

ieta

ry

inta

ke

Moe

eni e

t al

54P

YM

S;

STR

ON

Gki

ds;

STA

MP

Iran

26 D

ec 2

010

to 1

9 Ja

n 20

11

Tert

iary

pae

dia

tric

te

achi

ng h

osp

ital

119

1–17

yea

rsM

alig

nanc

y, L

OS

 <24

hou

rsW

HO

(WFH

, HFA

, BM

I if W

FH o

r H

FA m

issi

ng)

Moe

eni e

t al

55S

TRO

NG

kid

s-si

mp

lified

New

Zea

land

30 O

ct t

o 30

D

ec 2

012

Clin

ical

and

sur

gica

l p

atie

nts

162

1 m

onth

to

17 y

ears

LOS

 <24

hou

rs, o

ncol

ogic

al

pat

ient

sW

HO

(BM

I, W

FA, H

FA, W

FH)

Mur

phy

et

al39

SC

AN

Aus

tral

iaN

ot r

epor

ted

Pae

dia

tric

onc

olog

y p

atie

nts

325–

18 y

ears

LOS

 <24

hou

rs, c

linic

ally

un

stab

le, c

ond

ition

s th

at a

ffect

hy

dra

tion,

non

-Eng

lish

spea

king

SG

NA

Rub

et

al57

STA

MP

; STA

MP

-m

odifi

edIs

rael

Not

rep

orte

dO

utp

atie

nts

601–

6 ye

ars

Chi

ldre

n ca

red

for

in o

utp

atie

nt

clin

ics

Full

nutr

ition

al a

sses

smen

t:

die

tary

his

tory

, ant

hrop

omet

ric

mea

sure

men

ts, b

ody

com

pos

ition

ch

arac

teris

tics

Sou

ndar

ya e

t al

62C

AN

S s

core

Ind

iaN

ot r

epor

ted

Mat

erni

ty h

osp

ital

300

0–48

hou

rsN

ewb

orns

with

co

ngen

ital a

nom

alie

s,

new

bor

ns <

37 c

omp

lete

d

wee

ks g

esta

tion,

mul

tiple

p

regn

anci

es, N

ICU

car

e,

mot

her'

s ge

stat

iona

l dia

bet

es

mel

litus

, unr

elia

ble

est

imat

ion

of

gest

atio

nal a

ge

Ant

hrop

omet

ry b

ased

on

Ind

ian

grow

th c

hart

s (R

oher

’s P

ond

eral

In

dex

, hea

d c

ircum

fere

nce

to

leng

th r

atio

, che

st c

ircum

fere

nce

or M

UA

C a

nd/o

r M

UA

C t

o he

ad

circ

umfe

renc

e ra

tio, B

MI)

Sou

za D

osS

anto

s et

al40

Nut

ritio

n sc

reen

ing

tool

for

ped

iatr

ic p

atie

nts

with

 CF

Bra

zil

Not

rep

orte

dP

edia

tric

pat

ient

s w

ith

CF

dia

gnos

is82

6–18

yea

rsN

ot r

epor

ted

CFF

Con

sens

us R

epor

t cr

iteria

; N

utrit

iona

l ris

k sc

reen

ing

tool

p

rop

osed

by

McD

onal

d41

Sp

agnu

olo

et a

l58S

TRO

NG

kid

sIta

lyO

ct 2

012

to

Nov

201

2C

linic

al a

nd s

urgi

cal

pat

ient

s14

41–

18 y

ears

ICU

pat

ient

sW

HO

(BM

I) fo

r ac

ute

mal

nutr

ition

, W

HO

(HFA

) for

chr

onic

m

alnu

triti

on

Thom

as e

t al

65S

TAM

P; P

MS

T (S

TAM

P-

mod

ified

); P

YM

S

UK

Dec

201

4 to

M

arch

201

5C

hild

ren

in t

ertia

ry

hosp

ital a

cute

uni

ts30

00–

17.6

yea

rsN

ot r

epor

ted

.W

HO

(HFA

, WFH

, BM

I)

Tab

le 2

C

ontin

ued

Con

tinue

d

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

7Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Stu

dy

Too

lC

oun

try

Tim

e fr

ame

Po

pul

atio

nS

amp

le

size

Ag

e ra

nge

Exc

lusi

on

crit

eria

Ref

eren

ce s

tand

ard

Wan

g et

al6

STA

MP

Chi

naJu

ly 2

014

to

July

201

5In

pat

ient

s an

d

outp

atie

nts

with

SC

I45

1–12

yea

rsFe

ver

or d

iarr

hoea

on

adm

issi

on, h

isto

ry o

f he

pat

opat

hy, n

ephr

opat

hy o

r fo

rmal

die

tetic

ass

essm

ent

and

nu

triti

onal

the

rap

y th

at m

ay

have

intr

oduc

ed b

ias

tow

ard

s nu

triti

on s

tatu

s fr

om m

edic

al

reco

rds

WH

O (B

MI f

or a

ge, W

FA, H

FA)

Whi

te e

t al

9P

NS

T; S

GN

AA

ustr

alia

Sep

t 20

12 t

o Ju

ne 2

013

Clin

ical

and

sur

gica

l p

atie

nts

295

0–16

yea

rsLO

S <

24 h

ours

, clin

ical

ly

unst

able

, con

diti

ons

that

affe

ct

hyd

ratio

n, n

on-E

nglis

h sp

eaki

ng

WH

O (B

MI f

or a

ge, H

FA, W

FA)

for

0–2

year

s; C

DC

200

0 fo

r 2–

20 y

ears

; SG

NA

Wis

kin

et a

l59S

TAM

P;

STR

ON

Gki

ds;

SP

NR

S; P

YM

S

UK

Dec

200

9 to

Ju

ne 2

010

Pae

dia

tric

pat

ient

s w

ith

IBD

, out

pat

ient

s an

d

inp

atie

nts

463–

17 y

ears

Not

rep

orte

d.

ICD

-10;

mild

: wei

ght

SD

S <

-2;

mod

erat

e: S

DS

bet

wee

n −

2 an

d

−3;

sev

ere:

SD

S ≥

−3

Won

g et

al14

STA

MP

UK

Jan

to D

ec

2010

Pae

dia

tric

pat

ient

s w

ith

SC

I51

6 m

onth

s to

18

yea

rsLO

S <

24 h

ours

, IC

UFu

ll d

iete

tic a

sses

smen

t: c

linic

al,

nutr

ition

al, h

isto

rical

rec

ord

s

Won

oput

ri et

al56

PY

MS

; STA

MP

; S

TRO

NG

kid

s

Ind

ones

iaJa

n to

Feb

20

14C

linic

al p

atie

nts

116

1–15

yea

rsLO

S <

24 h

ours

SG

NA

, WH

O g

row

th c

hart

(res

ults

no

t re

por

ted

)

BM

I, b

ody

mas

s in

dex

; CA

NS

, Clin

ical

Ass

essm

ent

of N

utrit

iona

l Sta

tus;

CD

C, C

ente

rs fo

r D

isea

se C

ontr

ol a

nd P

reve

ntio

n; C

F, c

ystic

fib

rosi

s; C

FF, C

ystic

Fib

rosi

s Fo

und

atio

n; H

FA, h

eigh

t fo

r ag

e; IB

D, i

nflam

mat

ory

bow

el d

isea

se; I

CD

, Int

erna

tiona

l Sta

tistic

al C

lass

ifica

tion

of D

isea

ses

and

Rel

ated

Hea

lth P

rob

lem

s; IC

U, i

nten

sive

car

e un

it; L

OS

, len

gth

of h

osp

ital s

tay;

MA

C, m

id-

arm

circ

umfe

renc

e; M

UA

C, m

id-u

pp

er a

rm c

ircum

fere

nce;

NIC

U, N

eona

tal I

nten

sive

Car

e U

nit;

NN

ST,

Neo

nata

l Nut

ritio

n S

cree

ning

Too

l; P

NS

T, P

aed

iatr

ic N

utrit

ion

Scr

eeni

ng T

ool;

PY

MS

, P

aed

iatr

ic Y

orkh

ill M

alnu

triti

on S

core

; SC

AN

, Nut

ritio

n S

cree

ning

Too

l for

Chi

ldho

od C

ance

r; S

CI,

spin

al c

ord

inju

ries;

SD

S, S

D s

core

; SG

A, S

ubje

ctiv

e G

lob

al A

sses

smen

t; S

GN

A, S

ubje

ctiv

e G

lob

al N

utrit

iona

l Ass

essm

ent;

SP

NR

S S

imp

le P

aed

iatr

ic N

utrit

ion

Ris

k S

core

; STA

MP,

Scr

eeni

ng T

ool f

or t

he A

sses

smen

t of

Mal

nutr

ition

in P

aed

iatr

ics;

STR

ON

Gki

ds,

Scr

eeni

ng T

ool f

or R

isk

on N

utrit

iona

l Sta

tus

and

Gro

wth

; TS

FT, t

ricep

s sk

in fo

ld t

hick

ness

; WFA

, wei

ght

for

age;

WFH

, wei

ght

for

heig

ht.

Tab

le 2

C

ontin

ued

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

8 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

and/or vice versa was considered as high risk in seven studies,7 9 38 39 56 59 62 and as a possibility for bias in 15 studies.2 6 14 32 40–42 51 53 58 60–62 64 65 Similarly, non-blinded interpretation of the reference standard with regards to the index test results was considered as a possibility for bias in 17 studies.2 14 32 39–42 51–53 56 58 60–62 64 65 The informa-tion about the patient flow and timing was considered to be unclear in 15 studies,2 7 14 32 38 40 42 53 54 56 58–60 62 65 as the intervals between the index tests and the reference stan-dard measurements were not reported. In one study,62 the anthropometric measurements were used as the index test and compared with the reference standard CANS score, which in our opinion introduces possible bias.

The reference standard was the second key domain of concern regarding applicability. Murphy et al39 used the SGNA tool for the reference standard, which is an assessment method rather than a reference standard.7 Similarly, Wonoputri et al56 defined the WHO anthropo-metric grow chart as the reference standard; however, the results presented only showed the comparisons of PYMS, STAMP and STRONGkids with SGNA. As mentioned before, Soundarya et al62 used the CANS score as the reference standard; however, in our report we present the results as reported in the study and also as calculated from the reported data in an inverted manner. The inclu-sion or exclusion criteria were not clearly defined in five studies.7 41 56 58 65

Evaluation of the screening/assessment toolsThe diagnostic accuracy of the validated screening/assessment tools for the chosen reference standards in the selected studies is presented in table 3.

Nine studies (34.61%) did not report Se, Sp, PPV and NPV; however, the data reported enabled the calculation of these validation metrics. Additionally, two studies did not report PPV and NPV, which were subsequently calcu-lated by the authors of the present study. The calculated values are highlighted with dagger (†) in table 3. In one study, only the Se was reported, with no data for the calcu-lation of the other validation metrics.55

The sensitivities of the screening/assessment tools ranged from 20%63 to 100%.6 39 52 56 59 SCAN with onco-logical patients and STAMP and STRONGkids with clinical patients showed the best results versus SGNA in terms of Se.39 56 Additionally, STAMP performed with 100% Se versus the anthropometric measurements for inpatients and outpatients with SCI.6 STAMP, STRONGkids, SPNRS and PYMS obtained 100% Se versus International Statis-tical Classification of Diseases and Related Health Prob-lems (ICD)-10; however, the Sp was poor (0%), except for PYMS (53.5%).59

The specificities ranged from 0% (STAMP, STRONGkids, SPNRS vs ICD-10)59 to 96.4% (STRONGkids vs PYMS).39 The combination of Se and Sp was evaluated as good only in four studies7 40 61 63 according to classification outlined in table 1.

Relatively high NPVs were seen for most of the studies, which ranged from 0%58 to 100% (SCAN vs SGNA;

STAMP vs SGNA; STRONGkids vs SGNA; PYMS vs ICD-10; STAMP vs nutritional intervention; STRONGkids vs nutri-tional intervention).6 39 52 59 On the contrary, the observed PPVs were a lot lower in most of the studies; these started from 2.5% (STAMP vs TSFT)63 and reached as high as 100% (STAMP, STRONGkids, SPNRS, PYMS, among each other).59

Agreement between the nutritional screening/assess-ment tool and the reference standard or other screening tool was verified in 12 studies (46.1%).6 7 14 32 38 40 53 56 57 59 63 65

As presented in table 3, all of the abovementioned validation metrics differed greatly when different cutoff values were used. The studies also included different populations and different sample sizes, and therefore direct comparisons of the results are not possible.

Seven studies (26.9%)7 14 32 41 51 55 61 also reported interobserver agreements, which varied from 0.53 for PYMS completed by two dietitians compared with the nursing staff,7 to 0.9 for STAMP completed by dietitians and nursing staff.32 Only two studies reported intraob-server agreement, where there was substantial agreement with the kappa value of 0.6 for STRONGkids

51 and 0.6 for STAMP.14

DIsCussIOnThis section discusses the results and the main findings of the present study. Recommendations for new studies that focus on validation of nutritional screening and assess-ment tools are proposed.

Two RQs were proposed in the present study, as follows.

rQ1: What are the currently available screening and assessment tools for detecting malnutrition in hospitalised paediatric patients?Currently, there are 14 nutritional screening tools and 3 nutritional assessment tools. In this systematic review, we identified validation studies of eight nutritional screening tools (SCAN, Nutritional screening tool for paediatric patients with CF, Nutritional risk screening tool in CF, NNST, PYMS, STRONGkids, STAMP and PNST) and three malnutrition risk assessment tools (SGA, SGNA and CANS score). Six screening tools were not included in this systematic review (NRS, SPNRS, PeDiSMART, SANSI, Screening tool to predict malnutrition among children under 2 years old in Zambia and Nutrition screening for infants and young children with special health care needs: A Look at Your Child’s Nutrition), as the studies identified did not include validation of the screening tool.

It is important to emphasise that the nutritional screening and assessment tools were developed and vali-dated for different populations of children, which were mainly focused on age limits and different exclusion criteria in terms of the admission diagnosis/status or other chronic illnesses.

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

9Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Tab

le 3

S

ensi

tivity

 (Se)

, sp

ecifi

city

 (Sp

), p

ositi

ve a

nd n

egat

ive

pre

dic

tive

valu

es (P

PV

and

NP

V) a

nd a

gree

men

t of

stu

die

s in

clud

ed in

the

sys

tem

atic

rev

iew

Stu

dy

Too

lR

efer

ence

sta

ndar

dS

e (%

)S

p (%

)R

atin

gP

PV

(%)

NP

V (%

)In

ter-

/int

ra o

bse

rver

ag

reem

ent

(kap

pa)

Ag

reem

ent

wit

h re

fere

nce

stan

dar

d/

oth

er s

cree

ning

to

ols

Rat

ing

ka

pp

a

Cho

urd

akis

et

al63

PY

MS

WH

O B

MI <

-2S

D*

90.9

1†81

.97†

g21

.90†

99.3

9†

STA

MP

77.2

7†81

.21†

f18

.61†

98.4

7†

STR

ON

Gki

ds

45.4

5*91

.74†

p23

.44†

96.8

0†

PY

MS

MU

AC

<-2

SD

*66

.67†

75.1

9†f

7.55

†98

.67†

STA

MP

81.8

2†78

.31†

f 9.

89†

99.3

3†

STR

ON

Gki

ds

41.6

7†92

.80†

p

15.1

5†98

.10†

PY

MS

TSFT

<-2

SD

*80

.00†

75

.07†

f 4.

08†

99.6

5†

STA

MP

40.0

0†78

.09†

p

2.50

†98

.93†

STR

ON

Gki

ds

20.0

0†92

.78†

p

3.7

0†98

.82†

PY

MS

STA

MP

*58

.70†

88.3

7†f

59.3

1†88

.11†

STA

MP

: k=

0.47

(95%

CI 0

.42

to 0

.53)

m

STA

MP

STR

ON

Gki

ds*

77.3

0†84

.11†

f 36

.82†

96.8

7†S

TRO

NG

kid

s: k=

0.47

(95%

CI 0

.42

to 0

.53)

m

STR

ON

Gki

ds

PY

MS

*31

.40†

96

.54†

p

74

.51†

81

.38†

P

YM

S: k

=0.

35(9

5% C

I 0.2

8 to

0.4

2)f

Dur

akb

aşa

et a

l60S

TRO

NG

kid

sW

HO

Acu

te m

alnu

triti

on:

WFH

, BM

I‡

48.0

0†

65.7

7†

p

13.6

4†

91.8

2†

Chr

onic

mal

nutr

ition

: H

FA‡

52.1

7†

45.1

8†

p

6.82

† 96

.54†

Gal

era-

Mar

tinez

et

al61

STR

ON

Gki

ds

exp

erts

WH

O B

MI <

-2S

D*

37.5

0†

91.1

6†

p

13.6

4†

97.5

1†

Inte

rob

serv

er: k

=0.

72(9

5% C

I 0.6

3-0.

80)m

STR

ON

Gki

ds n

on-

exp

erts

37.5

0†

94.4

2†

p

20.0

0†

97.6

0†

STA

MP

exp

erts

87.5

0†

82.7

9†

g 15

.90†

99

.44†

In

tero

bse

rver

: k=

0.74

(95%

CI 0

.67-

0.81

)m

STA

MP

non

-ex

per

ts62

.50†

79

.53†

f

10.2

0†98

.27†

Ger

asim

idis

et

al7

PY

MS

cFu

ll d

iete

tic

asse

ssm

ent*

5992

f 47

95In

tero

bse

rver

: k=

0.53

(95%

CI 0

.38

to 0

.67)

bA

Rc : k

=0.

46(9

5% C

I 0.2

7 to

0.6

4)m

STA

MP

c : k=

0.47

(95%

CI 0

.34

to 0

.61)

m

SG

NA

c : k=

0.12

(95%

CI −

0.11

- −

0.34

)s

PY

MS

d85

87g

4498

AR

d: k

=0.

51(9

5% C

I 0.4

0 to

0.7

0)m

STA

MP

d81

78f

3197

STA

MP

d: k

=0.

34(9

5% C

I 0.2

0 to

0.5

0)f

SG

NA

d15

100

p

100

91S

GN

Ad: k

=0.

24(9

5% C

I 0.1

0 to

0.5

0)f

Con

tinue

d

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

10 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Stu

dy

Too

lR

efer

ence

sta

ndar

dS

e (%

)S

p (%

)R

atin

gP

PV

(%)

NP

V (%

)In

ter-

/int

ra o

bse

rver

ag

reem

ent

(kap

pa)

Ag

reem

ent

wit

h re

fere

nce

stan

dar

d/

oth

er s

cree

ning

to

ols

Rat

ing

ka

pp

a

Hul

st e

t al

2S

TRO

NG

kid

sW

HO

(HFA

, WFH

)‡

75.3

2†

41.4

7†

p

22.5

7†

88.1

3†

Huy

sent

ruyt

et 

al51

STR

ON

Gki

ds

WH

O (W

FH, H

FA)‡

71

.9e

49.1

e p

11

.9e

94.8

e In

tero

bse

rver

: k=

0.61

69f

48.4

fp

10

.4f

94.8

fIn

trao

bse

rver

: k=

0.66

John

son

et a

l42N

NS

TU

K-W

HO

gro

wth

cha

rt

(WFA

)89

.675

.1f

32.9

98.1

Ling

et

al52

STA

MP

Nut

ritio

nal

inte

rven

tion‡

10

0†

40†

p

41.9

4†

100†

STR

ON

Gki

ds

100†

§ 38

.89†

§ p

35

.29†

§ 10

0†§

Mah

dav

i et

al38

SG

AA

nthr

opom

etric

an

d b

ioch

emic

al

mea

sure

men

ts W

HO

(W

FA, H

FA, W

FH,

TSFT

and

ser

um

tran

sfer

rin)

88.2

445

.83

p

60.6

180

.49

AR

i : k=

0.33

6f

Măr

gine

an e

t al

53S

TRO

NG

kid

sW

HO

(WFH

, HFA

)‡

97.0

0†50

.29†

f 53

.30†

96.6

3†A

R: k

=0.

61su

Mod

ified

S

TRO

NG

kid

s

97.0

0†

65.5

0†f

62.1

8†97

.39†

AR

: k=

0.71

6su

Mar

tinez

-N

adal

et

al64

CA

N s

core

Roh

er’s

Pon

der

al in

dex

48.9

0†95

.47†

p

44.6

0†96

.16†

McC

arth

y et

al32

STA

MP

Full

nutr

ition

al

asse

ssm

ent*

70 (9

5% C

I 51

to 8

4)91

(95%

CI 8

6 to

94)

f 54

.8 (9

5% C

I 38

.8 t

o 69

.8)

94.9

(95%

CI

90.5

to

97.4

)In

tero

bse

rver

: k=

0.92

(95%

CI 0

.76

to 1

.00)

AR

: k=

0.54

(95%

CI 0

.46

to 0

.62)

m

McD

onal

d41

Nut

ritio

n R

isk

Scr

eeni

ngC

FF C

onse

nsus

R

epor

t cr

iteria

‡ 84

75f

82†

77†

Inte

rob

serv

er: k

=0.

85a

Tool

in C

FFu

ll d

iete

tic

asse

ssm

ent‡

86

78f

84†

80†

Moe

eni e

t al

54P

YM

SW

HO

(WFH

, HFA

, B

MI i

f WFH

or

HFA

m

issi

ng)‡

90.0

0†31

.46†

p

30.6

8†90

.32†

STR

ON

Gki

ds

83.3

3†49

.44†

p

35

.71†

89

.80†

STA

MP

90.0

0†

37.0

8†

p

32.5

3†

91.6

7†

Moe

eni e

t al

55S

TRO

NG

kid

sW

HO

(BM

I, W

FA, H

FA,

WFH

)84

k/

Inte

rob

serv

er: k

=0.

65j

sim

plifi

ed90

I /

Mur

phy

et

al39

SC

AN

SG

NA

100

(95%

CI 7

6 to

10

0)39

(95%

CI 1

7 to

64)

p

56 (9

5% C

I 35

to 7

6)10

0 (9

5% C

I 59

to 1

00)

Rub

et

al57

STA

MP

Full

nutr

ition

al

asse

ssm

ent‡

47

.6 (9

5% C

I 28.

3 to

67.

6)94

.9 (9

5% C

I 81

to 9

9)p

83

.377

.1A

R: k

=0.

47(9

5% C

I 0.3

5 to

0.7

9)m

STA

MP

-mod

ified

76.1

9 (9

5% C

I 54.

91

to 8

9.37

)82

.05

(95%

CI 6

7.33

to

91.

02)

f 69

.686

.5A

R: k

=0.

57(9

5% C

I 0.3

5 to

0.7

9)m

Sou

ndar

ya e

t al

62C

AN

Sco

reR

oher

’s P

ond

eral

ind

ex65

.286

.4f

60.2

88.7

60.2

6†

88.7

4†

f 65

.28†

86

.40†

Tab

le 3

C

ontin

ued

Con

tinue

d

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

11Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Stu

dy

Too

lR

efer

ence

sta

ndar

dS

e (%

)S

p (%

)R

atin

gP

PV

(%)

NP

V (%

)In

ter-

/int

ra o

bse

rver

ag

reem

ent

(kap

pa)

Ag

reem

ent

wit

h re

fere

nce

stan

dar

d/

oth

er s

cree

ning

to

ols

Rat

ing

ka

pp

a

BM

I84

.773

.6f

50.4

93.8

50.4

1†

93.8

5†

f 84

.72†

73

.68†

Roh

er’s

Pon

der

al

ind

ex+

BM

I84

.772

.8f

49.5

93.7

Mid

arm

ci

rcum

fere

nce/

head

41.6

77.6

p

37.0

80.8

circ

umfe

renc

e ra

tio37

.04†

80

.82†

p

41

.67†

77

.63†

Sou

za D

osS

anto

s et

al40

Nut

ritio

n sc

reen

ing

tool

for

ped

iatr

ic

pat

ient

s w

ith C

F

CFF

Con

sens

us

Rep

ort

crite

ria‡

72.4

71.7

f 58

.33†

82

.61†

A

R: k

=0.

418

(p<

0.00

1)m

Nut

ritio

nal r

isk

scre

enin

g to

ol

pro

pos

ed b

y M

cDon

ald

‡

8595

.2g

94.4

4†

86.9

6†

AR

: k=

0.80

4 (p

<0.

001)

su

Sp

agnu

olo

et a

l58S

TRO

NG

kid

sW

HO

(BM

I, H

FA fo

r ac

ute

and

chr

onic

m

alnu

triti

on)‡

71 (9

5% C

I 48

to 8

9)53

(95%

CI 4

3 to

63)

f 21

(95%

CI 1

7 to

25)

85 (9

5% C

I 85

to 9

0)

Thom

as e

t al

.65S

TAM

PW

HO

(HFA

, WFH

, B

MI)‡

63

.236

.3p

35

.663

.8A

R: k

=−

0.00

5n

PM

ST

(STA

MP

-m

odifi

ed)

94.4

29.0

p

40.5

91.1

AR

: k=

0.17

7s

PY

MS

26.1

67.1

p

34.3

58.0

AR

: k=

−0.

71n

Wan

g et

al6

STA

MP

WH

O (W

HA

, HFA

, B

MI)‡

10

073

.3f

65.2

100

AR

: k=

0.60

3m

Whi

te e

t al

9P

NS

TW

HO

and

CD

C 2

000

BM

I ≤ −

2SD

‡ 89

.366

.2f

22.5

98.4

WFA

≤ −

2SD

‡ 89

.565

.0f

15.3

98.9

HFA

≤ −

2SD

‡ 55

.662

.4f

4.5

97.8

SG

NA

‡ 77

.882

.1f

69.3

87.6

SG

NA

WH

O a

nd C

DC

200

0

BM

I ≤ −

2SD

‡ 96

.572

.5f

27.7

99.5

WFA

≤ −

2SD

‡ 85

.769

.7f

17.8

98.5

HFA

≤ −

2SD

‡ 46

.266

.5p

6.

096

.4

Wis

kin

et a

l59S

TAM

PIC

D-1

0‡

100†

0†

p

6.

52†

NA

† A

R: k

=−

0.01

4n

STR

ON

Gki

ds

100

0†

p

6.52

† N

A†

AR

: k=

−0.

013

n

SP

NR

S10

0†

0†

p

6.52

† N

A†

AR

: k=

−0.

013

n

PY

MS

100†

53

.49†

f

13.0

4†

100†

A

R: k

=0.

079

s

STA

MP

STR

ON

Gki

ds‡

10

0†

NA

† p

10

0†

NA

† S

TRO

NG

kid

s: k=

0.77

4su

STR

ON

Gki

ds

SP

NR

S‡

100†

N

A†

p

100†

N

A†

SP

NR

S: k

=0.

732

su

Tab

le 3

C

ontin

ued

Con

tinue

d

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

12 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

Stu

dy

Too

lR

efer

ence

sta

ndar

dS

e (%

)S

p (%

)R

atin

gP

PV

(%)

NP

V (%

)In

ter-

/int

ra o

bse

rver

ag

reem

ent

(kap

pa)

Ag

reem

ent

wit

h re

fere

nce

stan

dar

d/

oth

er s

cree

ning

to

ols

Rat

ing

ka

pp

a

PY

MS

‡ 50

† N

A†

p

100†

0†

PY

MS

: k=

0.33

2f

SP

NR

SS

PN

RS

‡ 10

0†

NA

†p

10

0†

NA

† S

PN

RS

: k=

0.60

0m

PY

MS

‡ 50

† N

A†

p

100†

0†

P

YM

S: k

=0.

270

f

PY

MS

‡ 50

† N

A†

p

100†

0†

PY

MS

: k=

0.23

6f

Won

g et

al14

STA

MP

Full

die

tetic

as

sess

men

t‡

83.3

66.7

f 78

.173

.6In

tero

bse

rver

g : k=

0.75

2A

R: k

=0.

507

(95%

CI 0

.646

to

1.00

0)m

PY

MS

‡ In

trao

bse

rver

h : k=

0.63

5P

YM

S: k

=0.

314

(95%

CI 0

.076

to

0.55

2)f

Won

oput

ri et

al56

PY

MS

SG

NA

‡ 95

.31

(95%

CI 0

.87

to 0

.98)

76.9

2 (9

5% C

I 63

to 8

6)f

83.5

6 (9

5% C

I 73

to 9

0)93

.02

(95%

CI

81 t

o 97

)A

Re : k

=0.

348

(95%

CI 0

.191

to

0.50

6)f

AR

f : k=

0.12

5(9

5% C

I 0 t

o 0.

299)

s

STA

MP

100

(95%

CI 0

.94

to 1

)11

.54

(95%

CI 5

to

23)

p

58.2

(95%

CI 4

8 to

67)

100

(95%

CI 6

1 to

100

)A

Re : k

=0.

018

(95%

CI 0

to

0.14

0)s

AR

f : k=

0(9

5% C

I 0 t

o 0.

140)

s

STR

ON

Gki

ds

100

(95%

CI 9

4 to

10

0)7.

7 (9

5% C

I 3 t

o 18

)p

57

.14

(95%

CI

47.9

to

65.9

)10

0 (9

5% C

I 51

to 1

00)

AR

e : k=

0.02

8(9

5% C

I 0 t

o 0.

149)

s

AR

f : k=

0(9

5% C

I 0 t

o 0.

144)

s

*low

and

med

ium

mal

nutr

ition

al r

isk

vers

us h

igh.

†not

rep

orte

d b

y au

thor

s (ie

, cal

cula

ted

in t

he p

rese

nt s

tud

y).

‡low

ver

sus

med

ium

and

hig

h m

alnu

triti

onal

ris

k.§e

rror

in r

epor

ted

dat

a.aa

mon

g si

x re

gist

ered

die

titia

ns o

n 18

pat

ient

s.b

two

rese

arch

die

titia

ns a

nd w

ard

nur

sing

sta

ff. 

cnur

se-r

ated

mal

nutr

ition

ris

k sc

reen

ing

tool

.d

rese

arch

-die

titia

n-ra

ted

mal

nutr

ition

ris

k sc

reen

ing

tool

. er

efer

ence

sta

ndar

d: a

cute

mal

nutr

ition

(WFH

<-2

SD

).fr

efer

ence

sta

ndar

d: c

hron

ic m

alnu

triti

on (H

FA <

-2S

D).

greg

iste

red

die

titia

n ve

rsus

nur

ses.

hreg

iste

red

die

titia

n.io

n 93

pat

ient

s.jp

aed

iatr

icia

n ve

rsus

nur

ses.

knur

se.

lpae

dia

tric

ian.

mre

gist

ered

die

titia

n/p

hysi

cian

sp

ecia

lised

in p

aed

iatr

ic n

utrit

ion

vers

us n

urse

non

-sp

ecia

lized

in n

utrit

ion

or p

aed

iatr

ic r

esid

ent.

AR

, agr

eem

ent

with

the

ref

eren

ce s

tand

ard

use

d; f

or o

ther

ab

bre

viat

ions

and

acr

onym

s, s

ee t

able

 1.

Tab

le 3

C

ontin

ued

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

13Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

rQ2: What is the validity of the screening and assessment tools versus the reference standard?This systematic review evaluated 26 studies in all. The methodological quality of all of these was considered moderate. The main methodological problems were related to the lack of detailed descriptions of the study protocols, and to the non-blinded interpretation of index tests with regards to the reference standards, and/or vice versa. In particular, the study by Soundarya et al62 was inadequate in the reporting of information about the study protocol. There was no information about the study flow, the time frame and the number of people involved in the evaluation process. Additionally, the CANS score was used as the reference standard and not as the index test, as would be expected. Apart from the abovemen-tioned study, four additional studies did not report the time frame of the study.39–41 57

Direct comparisons of the screening tools in terms of Se, Sp, PPV and NPV are not possible, as different cutoff values were used in the different studies. Also, the three malnutrition risk groups (ie, low, medium and high) were not uniquely combined into two groups for comparisons. Some studies used the combination of low to medium risk compared with high risk, while other studies used low compared with medium to high risk.

STAMP and STRONGkids were the most often validated screening tools in the investigated studies. Anthropom-etry measurements were used as the reference standard in five studies for validation of STAMP, and in nine studies for validation of STRONGkids. Clinical and surgical paedi-atric patients (n=223) were included in a study using BMI measurements as the reference standard, while comparing low and medium risk groups versus the high risk group.61 STAMP was validated with a good rating when performed by experts, but rated only fair with non-experts. Similar results were obtained on a larger sample in a multicenter study (n=2567) when validated with BMI and MUAC, but these were poor when validated with triceps skin fold thickness (TSFT).63 However, for comparisons of low risk versus medium and high risk groups, the validation studies reported poor ratings, with the exception of the study that included only a specifically small group of paediatric patients with SCI (n=45).6 When compared with full dietetic assessment, STAMP obtained fair ratings in three studies7 14 32 and poor ratings in the study on 60 outpatients.57 The modified version of STAMP in the last study here obtained a fair rating. Additionally, when vali-dated with SGNA56 and ICD-10,58 the ratings were poor.

STRONGkids obtained fair ratings when compared with anthropometric measurements in only two studies.53 58 The validation of STRONGkids on the same group of patients as STAMP resulted in lower agreement with the BMI measurements regardless of the expertise of the assessor.61 Similar conclusions can be drawn from the larger multicenter study on 2567 patients.63

When validated with SGNA, STRONGkids obtained the highest Se (100%); however, the Sp was very low (7.7%).56 Unfortunately, the results of the validation

with anthropometry were not reported. The authors report only the prevalence of malnutrition based on the WHO criteria. Although the number of patients included was not so small (n=116), the authors reported a high percentage of children with oncological disorders (43.1%) and infectious diseases (14.6%), which resulted in a higher percentage of positive samples (28.4%). In the same study, PYMS compared with SGNA obtained the best results (Se=95.3%, Sp=76.9%, PPV=83.6%, NPV=93.0%). These results deviated quite a lot compared with STAMP and STRONGkids in the same patient group. However, no conclusions can be drawn here as the results were reported only when compared with other screening/assessment tools, and not using an anthropometry measurement as the reference standard.56 Similar results were found in the study with only 46 patients with IBD.59

On the contrary, the CANS score was validated using anthropometry measurements (Roher’s Ponderal Index) on the largest group of newborns (n=14 477), and it gave a Sp of >90% and a Se of <50%.64 The CANS score showed a better performance compared with Roher’s Ponderal Index and BMI on a much smaller sample of 300 newborns.62 The authors of both studies concluded that the CANS score is useful for the identification of fetal malnutrition in newborns; however, the results from the larger sample did not confirm these statements. The second malnutrition screening tool for newborns, NNST, was also validated on a relatively large group of patients (n=909) using anthropometry measurements, where the UK growth charts obtained fair results (Se=89.6%, Sp=75.1%).42 However, the NNST was designed for a specific group of patients: neonatal patients in ICU who were >2 weeks old. The results showed good perfor-mance, although the low PPV indicates that two-thirds of the patients were unnecessarily wrongly predicted as at risk of malnutrition.

The PNST was validated in only one study with anthro-pometric measures (BMI ≤−2SD; Se=89.3%, Sp=66.2%).9 However, the SGNA was also validated with anthropo-metric measures (BMI≤−2SD) in the same study on the same group of patients. These results showed even better agreement (Se=96.5%, Sp=72.5%) although the very low PPVs indicated high overprediction of positive cases. The SGNA was also used as the reference standard for valida-tion of PNST, which obtained a fair performance.

The PYMS screening tool was validated with anthro-pometric measurements in three studies,54 63 65 and it obtained a good rating compared with BMI in a multi-country study that included 2567 paediatric patients,63 with a poor rating in the other two smaller studies (n=300, n=119). When compared with STAMP and STRONGkids, PYMS obtained the best results in the validation with BMI, for Se and Sp. Very low PPVs can be a source of concern, which was indeed noted by the authors of the tool in their first validation study using full dietetic assessments as the reference standard.7

When the PYMS, STAMP, STRONGkids and PNRS were validated with ICD-10 as the reference standard, they

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

14 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

showed poor agreement.59 This study included a small (n=46) and very heterogeneous group of children with IBD, and it did not reach any conclusions about the reasonability of the use of these tools for children with IBD.

The PYMS was also validated with the SGNA, which obtained a fair rating regarding the Se and Sp.56 High PPVs and NPVs also indicated that overprediction is not a cause for concern. As reported by the authors, PYMS obtained the best agreement with SGNA for the detection of acute malnutrition (k=0.3).

Three malnutrition screening tools specially designed for children with specific chronic illnesses were identi-fied in this study39–41 (SCAN, Nutrition Risk screening tool in CF, Nutrition screening tool for paediatric patients with CF). None of these tools were validated with anthropometry measurements. SCAN was validated with SGNA,39 and obtained a poor rating due to weak iden-tification of negative cases (Sp=39%); however, identifi-cation of positive cases was 100%. The PPV showed that >40% of the identified children were actually not exposed to the risk. The Nutrition Risk screening tool in CF was validated using full dietetic/nutritional assessment and the CFF consensus report criteria as the reference stan-dard, which obtained a fair rating with high PPVs and NPVs, thus indicating a good performance.41 The second tool for patients with CF (ie, Nutrition screening tool for paediatric patients with cystic fibrosis) was also validated with the CFF consensus report criteria and obtained fair agreement, but with lower PPVs and NPVs, which thus indicated more problems with the overprediction of posi-tive cases (>40%).40 Good agreement was reported when this tool was validated with the Nutrition Risk screening tool in CF (k=0.8).

Comparisons with similar systematic reviewsSimilar reviews of validation studies on paediatric malnutrition screening tools can be found in the litera-ture.8 66 67 However, the present systematic review includes the greatest number of paediatric malnutrition screening/assessment tools identified, and also in terms of the vali-dation studies. The systematic reviews by Moeeni and Day68 and Hartman et al69 only focused on the description of the available paediatric malnutrition screening tools (six and five tools, respectively). Six paediatric malnutri-tion screening tools were described in the study by Joosten and Hulst,8 with eight validation studies included. The authors defined two tools as the most practical and reli-able: STRONGkids and PYMS. They proposed that STRON-Gkids is the most reliable for assessing nutritional risk, and PYMS for assessing nutritional risk and actual nutritional status. A systematic review of studies that validated malnu-trition screening tools for hospitalised children and included a meta-analysis was published by Huysentruyt et al.66 The systematic review included four malnutrition screening tools (PYMS, STRONGkids, STAMP and PNRS) and 15 validation studies. As also observed in the present study, the authors were confronted with several problems

when comparing the validation results from several studies. This was in particular due to the heterogeneity of the reference standards, the different cutoff points used and the small sample sizes (only one study had more than 100 participants). Their conclusions demonstrated that at the time there was insufficient evidence to choose one screening tool over another. The most recent systematic review was conducted by Teixeira and Viana,67 and this included five malnutritional screening/assessment tools (PYMS, STRONGkids, STAMP, PNST and SGNA). The authors concluded that STRONGkids and STAMP showed the best clinical performances in the studies included.

The results of the present systematic review are not consistent with the conclusions of Josten and Hulst8 and Teixseira and Viana,67 who recommended STRONGkids as the most reliable screening tool; here, PYMS showed better performance (table 3).

strengths and limitationsThere are some key limitations to the present study that have to be emphasised. The most important limitation comes from the lack of a gold standard for evaluation of the malnutrition risk of hospitalised children. Conse-quently, the studies used different reference standards, most often as anthropometric measures or dietetic assess-ments. As reported in a number of studies,70–72 the accu-racy of the anthropometric measurements was often poor, which resulted in questionable uniformity of the valida-tion of these screening/assessment tools. Full dietetic/nutritional assessment also varied due to the different methods used and the different educational standards for dietetics in different countries.3 Some studies used another screening/assessment tool for the reference standard, which introduces a certain source of bias. As observed in the present study, the PPVs in the validation studies that compared one screening tool to another were higher than for the same screening tool compared with the anthropometry measures. However, this can be expected if both of the screening tools overpredict posi-tive cases when validated with the same anthropometry measurements as reference standard. The choice of the reference standard, therefore, represents a source of bias to the original studies, and consequently also to the present systematic review.

Another limitation comes from the different inclu-sion and exclusion criteria used in the studies, and the lower power of the studies with small sample sizes. The protocols of the studies were not uniquely defined, so the studies differed in the type and number of assessors using the nutritional screening tool studied. Similarly, different types and numbers of assessors were involved in evalua-tion of the malnutrition risk, according to the study-de-fined reference standard. Also, almost one-third of the studies evaluated (30.8%)6 14 39–41 52 57 59 had a sample size of <100 patients. These small numbers of paediatric patients involved mandate caution when generalising the results to the full population.

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

15Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

However, one of the main strengths of the present systematic review is that we focused only on studies that included validation of the screening/assessment tools used based on the chosen reference standard. We are confident that all of the currently available nutritional screening/assessment tools for paediatric patients are included in this review. This study thus presents a complete review of the success of these tools in the prediction of malnutrition risk.

recommendationsThe results of this study show the need for the defi-nition of a gold standard. We propose that an expert group is formed to discuss and define which reference standard should be used worldwide for the evaluation of screening tools. Based on the deficiencies identified in validation studies during the present systematic review, we recommend that the reference standard should never be another screening/assessment tool. We propose that the anthropometric measurements defined in the WHO growth charts/Anthro/ AnthroPlus software or the CDC 2000 growth charts/Epi Info 7 should be considered as the basic reference standard for the purpose of fair comparisons. The American Society for Parenteral and Enteral Nutrition (ASPEN)73 recommendations included the recording of weight, height, BMI and MUAC, and considered the TSFT and mid-arm muscle circumference on admission, with reference to the appropriate growth chart. Head circumference must also be obtained in infants younger than <2 years. The CDC74 and ASPEN73 recommend using the WHO charts for children up to 2 years of age. For children and adolescents from 2 to 20 years of age, they recommend using the CDC 2000 charts. The newest versions of growth charts should also be used. Anthropometric measurements should be performed using calibrated equipment, according to the examination protocols described by the CDC.75 As validation of malnu-trition screening tools using anthropometry measure-ments as the reference standard tend to produce a lot of false-positive results, our recommendation is to use full dietetic/nutritional assessment in the second stage only for the positively identified cases. However, the dietetic/nutritional assessment process and evaluation should also be standardised first. Furthermore, specific disease condi-tions can cause energy and/or protein imbalances, and therefore these should be considered in the final evalua-tion of the nutritional status.73 Validation studies should also test several malnutrition screening tools on the same population, to avoid bias due to different patient popula-tions, disease backgrounds or age groups.31

It is important that healthcare professionals who perform nutritional screening are appropriately educated and trained in the measurement of the anthropometric parameters, and they should use the appropriate growth charts or computer software, and the chosen screening tool.

The study protocol should be carefully designed and followed, with particular attention paid to the following:

► Patient selection: The sample of patients included in studies should be as large as possible, with consecu-tive or random sampling used, and inappropriate exclusions avoided. The inclusion and exclusion criteria should be carefully defined. All of the patients included in a study should be included in the valida-tion procedures.

► Flow and timing: To obtain an evaluation of the nutritional status that is as objective as possible, the evaluator using the screening test and the reference standard should not be the same person. Additionally, at the time of performing the evaluation of the nutri-tional status, the evaluators should not be acquainted with the results of other evaluations. The exact sequence of the performing of the screening tests and the reference standards has to be defined in the study protocol, while taking into consideration that both the screening test and the reference standard should be evaluated on admission73 and on the same day (or as close as possible). This will avoid possible changes in the patient health condition in between these eval-uations. The exact flow and timing of these should also be reported.

► Reporting results: All of the traditional evaluation metrics should be reported, such as Se, Sp, PPV and NPV. When the malnutrition risk is evaluated using three categories (ie, low/moderate/high), the moderate and high risks should both be treated as ‘at risk’. A table showing the cross-classification of the malnutrition risk on the screening test compared with the reference standard should also be reported. When several evaluators use the screening test or the reference standard, the inter-rater or intra-rater agreement should also be reported.

COnClusIOnsThis systematic review shows that several paediatric nutritional screening/assessment tools have been developed; however, due to the lack of a gold stan-dard, it is very difficult (if not impossible) to compare them at present. The validation results show that nutri-tional screening tools perform better when designed for specific groups of patients who suffer from chronic or specific conditions. An exception is seen for SCAN, which is designed for oncological patients. The only validation study of SCAN that was found for inclusion in the present systematic review used SGNA as the refer-ence standard; therefore, additional validation studies are needed for correct validation here. Low PPVs were seen for almost all of the studies that used anthropom-etry as the reference standard, which indicates the prob-lems associated with overprediction of positive cases. It is true that it is better to include more false positives than false negatives, but this also leads to unnecessary exposure of the children to more invasive assessments, an increased workload for the health staff and an addi-tional financial burden. However, very low PPVs should be treated with caution.

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

16 Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

It is particularly difficult to recommend any one screening/assessment tool on the basis of the results of all of these published studies, due to their heterogeneity. However, PYMS appears to perform better than STAMP and STRONGkids when compared with anthropometric measurements, especially in terms of BMI and TSFT. Therefore, we would recommend the use of PYMS in the hospital setting for paediatric patients without chronic conditions. Due to its tendency to overpredict positive cases, we also recommend the use of full dietetic/nutri-tional assessments in the second stage for the positively identified cases.

For fair comparisons here, there is the need for more studies that are aimed at the validation of different screening/assessment tools for the same group of patients using the same reference standard. We also recommend that a unified standard for full nutritional assessment should be developed, and that this should then be used in combination with the cited growth charts.

Thus, we recommend further studies to validate nutri-tional screening/assessment tools with the aim being to provide health experts with fair comparisons, and conse-quently easier decisions, in terms of which tool(s) to use.

Contributors PK and PPB conceived the study design. PK, PPB, NMV and MP performed the data extraction and analysis and performed the systematic review. All of the authors have read and approved the final version of the manuscript.

Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests None declared.

Patient consent for publication Not required.

Provenance and peer review Not commissioned; externally peer reviewed.

Data sharing statement All of the data were collected from previously published studies. Our dataset is available from the corresponding author on request.

Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http:// creativecommons. org/ licenses/ by- nc/ 4. 0/.

rEFErEnCEs 1. Campanozzi A, Russo M, Catucci A, et al. Hospital-acquired

malnutrition in children with mild clinical conditions. Nutrition 2009;25:540–7.

2. Hulst JM, Zwart H, Hop WC, et al. Dutch national survey to test the STRONGkids nutritional risk screening tool in hospitalized children. Clin Nutr 2010;29:106–11.

3. Joosten KF, Hulst JM. Prevalence of malnutrition in pediatric hospital patients. Curr Opin Pediatr 2008;20:590–6.

4. Okoromah CA, Ekure EN, Lesi FE, et al. Prevalence, profile and predictors of malnutrition in children with congenital heart defects: a case-control observational study. Arch Dis Child 2011;96:354–60.

5. Pawellek I, Dokoupil K, Koletzko B. Prevalence of malnutrition in paediatric hospital patients. Clin Nutr 2008;27:72–6.

6. Wang YJ, Zhou HJ, Liu PJ, et al. Risks of undernutrition and malnutrition in hospitalized pediatric patients with spinal cord injury. Spinal Cord 2017;55:247–54.

7. Gerasimidis K, Keane O, Macleod I, et al. A four-stage evaluation of the Paediatric Yorkhill Malnutrition Score in a tertiary paediatric hospital and a district general hospital. Br J Nutr 2010;104:751–6.

8. Joosten KF, Hulst JM. Nutritional screening tools for hospitalized children: methodological considerations. Clin Nutr 2014;33:1–5.

9. White M, Lawson K, Ramsey R, et al. Simple nutrition screening tool for pediatric inpatients. JPEN J Parenter Enteral Nutr 2016;40:392–8.

10. Elia M, Stratton RJ. Considerations for screening tool selection and role of predictive and concurrent validity. Curr Opin Clin Nutr Metab Care 2011;14:425–33.

11. Elia M, Stratton RJ. An analytic appraisal of nutrition screening tools supported by original data with particular reference to age. Nutrition 2012;28:477–94.

12. Erkan T. Methods to evaluate the nutrition risk in hospitalized patients. Turk Pediatri Ars 2014;49:276–81.

13. Secker DJ, Jeejeebhoy KN. Subjective global nutritional assessment for children. Am J Clin Nutr 2007;85:1083–9.

14. Wong S, Graham A, Hirani SP, et al. Validation of the Screening Tool for the Assessment of Malnutrition in Paediatrics (STAMP) in patients with spinal cord injuries (SCIs). Spinal Cord 2013;51:424–9.

15. WHO. Growth reference data for 5-19 years. 2007 http://www. who. int/ growthref/ en/ (Accessed 4 Jan 2018).

16. WHO. The WHO Child Growth Standards, Documentation. 2016 http://www. who. int/ childgrowth/ standards/ en/ (Accessed 4 Jan 2018).

17. Centers for Disease Control and Prevention. CDC Growth Charts. 2016 https://www. cdc. gov/ growthcharts/ cdc_ charts. htm (Accessed 2 Feb 2018).

18. Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC Growth Charts for the United States: methods and development. Vital Health Stat 11 2002;11:1–190.

19. de Onis M, Onyango AW, Borghi E, et al. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ 2007;85:660–7.

20. Centers for Disease Control and Prevention. WHO Growth Standards Are Recommended for Use in the U.S. for Infants and Children 0 to 2 Years of Age. 2010 https://www. cdc. gov/ growthcharts/ who_ charts. htm (Accessed 4 Jan 2018).

21. de Onis M. Update on the implementation of the WHO child growth standards. World Rev Nutr Diet 2013;106:75–82.

22. Zong XN, Li H. Construction of a new growth references for China based on urban chinese children: Comparison with the WHO growth standards. PLoS One 2013;8:e59569.

23. Aburawi EH, Nagelkerke N, Deeb A, et al. National growth charts for United Arab Emirates children with Down syndrome from birth to 15 years of age. J Epidemiol 2015;25:20–9.

24. The British Dietetic Association. Model and Process for Nutrition and Dietetic Practice. Birmingham: The British Dietetic Association, 2016. https://www. bda. uk. com/ publications/ professional/ model_ and_ process_ for_ nutrition_ and_ dietetic_ practice_. (Accessed 23 Dec 2018).

25. WHO. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. WHO Technical Report Series No. 854. : World Health Organ Tech Rep Ser, 1995:854: 1–452. http:// apps. who. int/ iris/ bitstream/ 10665/ 37003/ 1/ WHO_ TRS_ 854. pdf. (Accessed 08 Mar 2018).

26. Cogill B. Anthropometric Indicators Measurement Guide. 2003 https://www. fantaproject. org/ sites/ default/ files/ resources/ anthropometry- 2003- ENG. pdf (Accessed 22 Apr 2018).

27. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Int J Surg 2010;8:336–41.

28. Review Manager (RevMan) [Computer program]. Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014. https:// community. cochrane. org/ help/ tools- and- software/ revman- 5. (Accessed 1 Oct 2017).

29. Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155:529–36.

30. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159–74.

31. van Bokhorst-de van der Schueren MA, Guaitoli PR, Jansma EP, et al. Nutrition screening tools: does one size fit all? A systematic review of screening tools for the hospital setting. Clin Nutr 2014;33:39–58.

32. McCarthy H, Dixon M, Crabtree I, et al. The development and evaluation of the Screening Tool for the Assessment of Malnutrition in Paediatrics (STAMP©) for use by healthcare staff. J Hum Nutr Diet 2012;25:311–8.

33. McCarthy H, McNulty H, Dixon M, et al. Screening for nutrition risk in children: the validation of a new tool. Journal of Human Nutrition and Dietetics 2008;21:395–6.

34. Gerasimidis K, Macleod I, Maclean A, et al. Performance of the novel Paediatric Yorkhill Malnutrition Score (PYMS) in hospital practice. Clin Nutr 2011;30:430–5.

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from

17Klanjsek P, et al. BMJ Open 2019;9:e025444. doi:10.1136/bmjopen-2018-025444

Open access

35. Baker JP, Detsky AS, Wesson DE, et al. Nutritional assessment: a comparison of clinical judgement and objective measurements. N Engl J Med 1982;306:969–72.

36. Detsky AS, Baker JP, Mendelson RA, et al. Evaluating the accuracy of nutritional assessment techniques applied to hospitalized patients: methodology and comparisons. JPEN J Parenter Enteral Nutr 1984;8:153–9.

37. Detsky AS, McLaughlin JR, Baker JP, et al. What is subjective global assessment of nutritional status?. JPEN J Parenter Enteral Nutr 1987;11:8–13.

38. Mahdavi AM, Ostadrahimi A, Safaiyan A. Subjective global assessment of nutritional status in children. Matern Child Nutr 2010;6:374–81.

39. Murphy AJ, White M, Viani K, et al. Evaluation of the nutrition screening tool for childhood cancer (SCAN). Clin Nutr 2016;35:219–24.

40. Souza Dos Santos Simon MI, Forte GC, da Silva Pereira J, et al. Validation of a Nutrition Screening Tool for Pediatric Patients with Cystic Fibrosis. J Acad Nutr Diet 2016;116:813–8.

41. McDonald CM. Validation of a nutrition risk screening tool for children and adolescents with cystic fibrosis ages 2–20 years. J Pediatr Gastroenterol Nutr 2008;6:438–46.

42. Johnson MJ, Pearson F, Emm A, et al. Developing a new screening tool for nutritional risk in neonatal intensive care. Acta Paediatr 2015;104:e90–e93.

43. Metcoff J. Clinical assessment of nutritional status at birth. Fetal malnutrition and SGA are not synonymous. Pediatr Clin North Am 1994;41:875–91.

44. Reilly HM, Martineau JK, Moran A, et al. Nutritional screening-evaluation and implementation of a simple Nutrition Risk Score. Clin Nutr 1995;14:269–73.

45. Sermet-Gaudelus I, Poisson-Salomon AS, Colomb V, et al. Simple pediatric nutritional risk score to identify children at risk of malnutrition. Am J Clin Nutr 2000;72:64–70.

46. Apostolou A, Printza N, Karagiozoglou-Lampoudi T, et al. Nutrition assessment of children with advanced stages of chronic kidney disease-A single center study. Hippokratia 2014;18:212–6.

47. Karagiozoglou-Lampoudi T, Daskalou E, Lampoudis D, et al. Computer-based malnutrition risk calculation may enhance the ability to identify pediatric patients at malnutrition-related risk for unfavorable outcome. JPEN J Parenter Enteral Nutr 2015;39:418–25.

48. Rowell A, Long C, Chance L, et al. Identification of nutritional risk by nursing staff in secure psychiatric settings: reliability and validity of St Andrew's Nutrition Screening Instrument. J Psychiatr Ment Health Nurs 2012;19:722–8.

49. Hasegawa J, Ito YM, Yamauchi T. Development of a screening tool to predict malnutrition among children under two years old in Zambia. Glob Health Action 2017;10:1339981.

50. Gilliam J, Laney SO. Nutrition screening for infants and young children with special health care needs: Spoken country, Washington. 2008 http://www. doh. wa. gov/ Portals/ 1/ Documents/ Pubs/ 970- 116_ Nutr itio nScr eeni ngFo rInf ants AndY oung CSHCN. pdf (Accessed 25 Jan 2018).

51. Huysentruyt K, Alliet P, Muyshont L, et al. The STRONG(kids) nutritional

screening tool in hospitalized children: a validation study. Nutrition 2013;29:1356–61.

52. Ling RE, Hedges V, Sullivan PB. Nutritional risk in hospitalised children: An assessment of two instruments. E Spen Eur E J Clin Nutr Metab 2011;6:e153–e157.

53. Mărginean O, Pitea AM, Voidăzan S, et al. Prevalence and assessment of malnutrition risk among hospitalized children in Romania. J Health Popul Nutr 2014;32:97–102.

54. Moeeni V, Walls T, Day AS. Assessment of nutritional status and nutritional risk in hospitalized Iranian children. Acta Paediatr 2012;101:e446–e451.

55. Moeeni V, Walls T, Day AS. The STRONGkids nutritional risk screening tool can be used by paediatric nurses to identify hospitalised children at risk. Acta Paediatr 2014;103:e528–e531.

56. Wonoputri N, Djais JT, Rosalina I. Validity of nutritional screening tools for hospitalized children. J Nutr Metab 2014;2014:1–6.

57. Rub G, Marderfeld L, Poraz I, et al. Validation of a nutritional screening tool for ambulatory use in pediatrics. J Pediatr Gastroenterol Nutr 2016;62:771–5.

58. Spagnuolo MI, Liguoro I, Chiatto F, et al. Application of a score system to evaluate the risk of malnutrition in a multiple hospital setting. Ital J Pediatr 2013;39:81.

59. Wiskin AE, Owens DR, Cornelius VR, et al. Paediatric nutrition risk scores in clinical practice: children with inflammatory bowel disease. J Hum Nutr Diet 2012;25:319–22.

60. Durakbaşa ÇU, Fettahoğlu S, Bayar A, et al. The prevalence of malnutrition and effectiveness of strongkids tool in the identification of malnutrition risks among pediatric surgical patients. Balkan Med J 2014;31:313–21.

61. Galera-Martínez R, Moráis-López A, Rivero de la Rosa MD, et al. Reproducibility and inter-rater reliability of 2 paediatric nutritional screening tools. J Pediatr Gastroenterol Nutr 2017;64:e65–e70.

62. Soundarya M, Basavaprabhu A, Raghuveera K, et al. Comparative assessment of fetal malnutrition by anthropometry and can score. Iran J Pediatr 2012;22:70–6.

63. Chourdakis M, Hecht C, Gerasimidis K, et al. Malnutrition risk in hospitalized children: use of 3 screening tools in a large European population. Am J Clin Nutr 2016;103:1301–10.

64. Martínez-Nadal S, Demestre X, Raspall F, et al. [Assessment of foetal nutrition status at birth using the CANS score]. An Pediatr 2016;84:218–23.

65. Thomas PC, Marino LV, Williams SA, et al. Outcome of nutritional screening in the acute paediatric setting. Arch Dis Child 2016;101:1119–24.

66. Huysentruyt K, Devreker T, Dejonckheere J, et al. Accuracy of nutritional screening tools in assessing the risk of undernutrition in hospitalized children. J Pediatr Gastroenterol Nutr 2015;61:159–66.

67. Teixeira AF, Viana KD. Nutritional screening in hospitalized pediatric patients: a systematic review. J Pediatr 2016;92:343–52.

68. Moeeni V, Day AS. Nutritional risk screening tools in hospitalised children. Int J Child Health Nutr 2012;1:39–43.

69. Hartman C, Shamir R, Hecht C, et al. Malnutrition screening tools for hospitalized children. Curr Opin Clin Nutr Metab Care 2012;15:303–9.

70. Geeta A, Jamaiyah H, Safiza MN, et al. Reliability, technical error of measurements and validity of instruments for nutritional status assessment of adults in Malaysia. Singapore Med J 2009;50:1013–8.

71. Moon RJ, Wilson P, Kirkham FJ, et al. Growth monitoring following traumatic brain injury. Arch Dis Child 2009;94:699–701.

72. Sullivan PB. Malnutrition in hospitalised children. Arch Dis Child 2010;95:489–90.

73. Mehta NM, Corkins MR, Lyman B, et al. Defining pediatric malnutrition: a paradigm shift toward etiology-related definitions. JPEN J Parenter Enteral Nutr 2013;37:460–81.

74. Grummer-Strawn LM, Reinold CM, Krebs NF, et al. Use of world health organization and cdc growth charts for children aged 0-59 months in the united states. In: Department of health and human services, centers for disease control and prevention. 2010. Centers for Disease Control and Prevention (2012) https://www. cdc. gov/ mmwr/ preview/ mmwrhtml/ rr5909a1. htm (Accessed 5 Feb 2018).

75. National Health and Nutrition Examination Survey (NHANES) anthropometry procedures manual. USA: CDC, 2009. https://www. cdc. gov/ nchs/ data/ nhanes/ nhanes_ 09_ 10/ lab. pdf. (Accessed 5 Feb 2018).

on January 24, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2018-025444 on 27 May 2019. D

ownloaded from